You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
ColossalAI/colossalai
Frank Lee 0e4e62d30d
[tensor] added __repr__ to spec (#1147)
2 years ago
..
amp [amp] included dict for type casting of model output (#1102) 2 years ago
builder [pipeline] refactor the pipeline module (#1087) 3 years ago
cli [hotfix] fix some bugs caused by size mismatch. (#1011) 3 years ago
communication [pipeline]refactor ppschedule to support tensor list (#1050) 3 years ago
context [usability] improved error messages in the context module (#856) 3 years ago
engine [pipeline]support more flexible pipeline (#1138) 2 years ago
fx [fx]add autoparallel passes (#1121) 2 years ago
gemini [gemini] gemini mgr supports "cpu" placement policy (#1118) 2 years ago
kernel [optim] refactor fused sgd (#1134) 2 years ago
logging [doc] improved docstring in the logging module (#861) 3 years ago
nn embedding op use gather_out (#1143) 2 years ago
pipeline [pipeline]add customized policy (#1139) 2 years ago
registry Remove duplication registry (#1078) 3 years ago
tensor [tensor] added __repr__ to spec (#1147) 2 years ago
testing [test] skip tests when not enough GPUs are detected (#1090) 3 years ago
trainer fix issue #1080 (#1071) 3 years ago
utils [zero] added error message to handle on-the-fly import of torch Module class (#1135) 2 years ago
zero [hotfix] prevent nested ZeRO (#1140) 2 years ago
__init__.py [NFC] polish __init__.py code style (#965) 3 years ago
constants.py fix typo in constants (#1027) 3 years ago
core.py [polish] polish singleton and global context (#500) 3 years ago
global_variables.py [MOE] add unitest for MOE experts layout, gradient handler and kernel (#469) 3 years ago
initialize.py [ddp] supported customized torch ddp configuration (#1123) 2 years ago